90 Days, 500 Bases, No Costed Plan: The Pentagon's FORTRESS Mandate Is a Duration Play for Switchgear, Not Reactors insight cover
Markets / EventCAT · CMI · GNRC•18 min read

90 Days, 500 Bases, No Costed Plan: The Pentagon's FORTRESS Mandate Is a Duration Play for Switchgear, Not Reactors

The Pentagon's FORTRESS initiative, unveiled by Defense Secretary Pete Hegseth on Sept. 30, 2026 and confirmed by chief spokesman Sean Parnell on Oct. 25, would require every major continental U.S. installation to run 90 days with no outside power, water, fuel or communications — with officials assessing roughly 500 bases and costs still unquantified. The investable read is duration, not capacity: autonomy goes from 14 days to 90, a 6.4x jump at unchanged critical load, which pushes incremental dollars into generators, batteries, switchgear and fuel storage rather than turbines or reactors. That points at the same OEM lines already supplying hyperscale, and at their cheaper multiples: Cummins at 15.2x trailing EV/EBITDA and Caterpillar at 17.7x, against Vertiv at 31.2x and Bloom Energy at 97.3x. The catch is timing — a second, appropriations-gated buyer rather than a budget-insensitive one, which makes the first published per-base cost the thing to watch.

Published Oct 30, 2026Updated Oct 30, 2026

Required off-grid autonomy

90 days

Fox News Digital, Oct 25, 2026; confirmed by Pentagon chief spokesman Sean Parnell the same day

Continental bases under assessment

~500

Senior defense official, Defense Writers Group roundtable, Oct 20, 2026

Duration multiple against today's standard

6.4x

90 days against the Army Reserve goal of 14 days without commercial power, Oct 22, 2026

Eaton Electrical Americas backlog

$15.2B

Up 33% from $11.4B a year earlier; Eaton 10-Q, quarter ended Jun 30, 2026

Caterpillar and Cummins already sell the same behind-the-meter hardware that hyperscale data centers buy. Now the Pentagon wants that equipment on roughly 500 domestic bases, each able to run 90 days without the grid. The question that matters for investors is not whether FORTRESS is real — it is which line items 76 extra days of autonomy actually buys.

Required off-grid autonomy

90 days

Fox News Digital, Oct 25, 2026; confirmed by Pentagon chief spokesman Sean Parnell the same day

Continental bases under assessment

~500

Senior defense official, Defense Writers Group roundtable, Oct 20, 2026

Duration multiple against today's standard

6.4x

90 days against the Army Reserve goal of 14 days without commercial power, Oct 22, 2026

Eaton Electrical Americas backlog

$15.2B

Up 33% from $11.4B a year earlier; Eaton 10-Q, quarter ended Jun 30, 2026

Cummins Power Systems segment EBITDA margin

48.9%

First half of 2026, up from 42.5%; Cummins 10-Q, Q2 2026

Time to full compliance, per officials

Decades

Air & Space Forces Magazine, Oct 20, 2026

Verified facts

What Was Announced, and What the Public Record Actually Supports

Defense Secretary Pete Hegseth unveiled FORTRESS — Federal Operations for Resilience, Troop Resources, Energy, Supply, and Survivability — in his Sept. 30, 2026 State of the Force address. Fox News Digital reported on Oct. 25, 2026, citing internal planning documents, that every major installation in the continental U.S. would be expected to sustain essential operations independently for at least 90 days, covering electricity, water, fuel and communications. Chief spokesman Sean Parnell confirmed the reporting and said mission-critical power must be \"independent, hardened, and redundant.\"

  • Confirmed: a 90-day off-grid standard for major continental installations, publicly confirmed by the Pentagon's chief spokesman on Oct. 25, 2026.
  • Confirmed: portable nuclear reactors, large-scale battery storage and islandable microgrids as the named technologies, plus strategic reserves of munitions, fuel, critical minerals, electronics, spare parts and medical supplies.
  • Confirmed: the Civil Reserve Manufacturing Network is folded in, with a requirement for at least two qualified supply pathways per critical dependency.
  • Not disclosed: the cost. A senior defense official said on Oct. 20 that officials are still trying to determine what this will cost, and that full compliance could take decades.
  • Not disclosed: a funded program line. No costed, appropriated FORTRESS budget appears in the public record.
Treat FORTRESS as a planning posture, not an order book: the 90-day standard is documented and confirmed, but the department has published no costed program and no timetable shorter than decades.

The mechanism

The Mandate Changes Duration, Not Capacity

The standard being replaced is a duration standard. The Army Reserve's published goal is to maintain mission capability for up to 14 days without commercial power. FORTRESS asks for 90. That is a 6.4x increase in the number of days a base must hold itself, with the load it carries broadly unchanged — and that arithmetic decides what gets bought.

Today's standard against the FORTRESS requirement
DimensionStandard todayFORTRESS requirement
Autonomy without utility power14 days (Army Reserve goal, Oct 22, 2026)90 days at every major continental installation
Installations in scope16 bases scheduled for microgrids by fiscal 2030 (2 Space Force, 14 Air Force)About 500 continental bases under assessment
Cost reference pointAbout $5B to rebuild Tyndall AFB after Hurricane Michael (2018)\"Scores of billions, if not hundreds of billions\" — costs still being assessed
Delivery timetableInstallations scheduled out to fiscal 2030Full compliance potentially decades away, per officials
Named technology setBackup generation and microgridsPortable nuclear reactors, large-scale battery storage, islandable microgrids, fuel and water stockpiles

Because critical load does not grow, neither does the generator fleet. What grows is everything that bridges time: on-site fuel storage sized for three months of consumption, batteries for black start and peak management, switchgear and protection to island the site, and controls to run it all without the utility. At \"scores of billions\" spread across 500 bases, implied spending is on the order of $100 million per base — a low-single-digit percentage of the $5B Tyndall rebuild. FORTRESS is therefore a standardized equipment program, not a construction program, and the money sits with equipment makers rather than contractors.

Who is paying now

The Growth on the Tape Is the Data-Center Buyer, Not the Defense Buyer

Latest reported revenue growth, year over year

Most recent reported quarter for each line. Segment figures where a segment is the cleanest read. The defense buyer appears in none of them.

Unit: %

Bloom Energy, Q3 2026

Company revenue

138.1%

Vertiv, Americas segment, Q3 2026

Segment net sales

74.5%

Generac, C&I segment, Q2 2026

Segment net sales

29.2%

Eaton, Electrical Americas, Q2 2026

Segment net sales

17.9%

Caterpillar, Q3 2026

Company revenue

17.3%

Cummins, Power Systems, Q2 2026

Segment external sales

15.5%

Powell Industries, quarter ended Jun 30, 2026

Company revenue

8.9%

None of that is FORTRESS. Eaton's Electrical Americas backlog reached $15,175 million at June 30, 2026, up 33% from $11,377 million a year earlier, with organic orders up 41% and a book-to-bill of 1.3 — and the segment attributes its organic growth to data center and machine OEM demand. Cummins says the same in its June-quarter filing: worldwide revenue rose on higher demand for power generation equipment, \"especially in data center applications.\" Generac reorganized into Residential and Commercial & Industrial segments in March 2026; C&I sales rose 29.2% year over year in the June quarter while Residential fell 2.2%.

The defense buyer arrives on an appropriations timetable, not a commercial one. No 2026 order book contains a single FORTRESS dollar, and real dollars appear only once a costed plan and a funded line exist.

Funding

The Second Buyer Is Appropriations-Gated, Not Budget-Insensitive

Calling FORTRESS a mandated order book overstates it, because the money is not committed. The statutory machinery already exists and is unglamorous. 10 U.S.C. § 2911 lets the department credit on-site generation that avoids backup power cost in its procurement cost-benefit analysis, requires an energy resilience master plan for each installation, requires the secretary to report the amount of critical energy load and the availability level deemed necessary by fiscal year, and mandates at least four military construction projects demonstrating on-site renewables serving critical-mission facilities.

  • Near-term funding runs through military construction and energy-resilience appropriations, with at least four demonstration projects already mandated by statute.
  • The Oct. 20 briefing pointed to public-private partnerships to carry capital-intensive costs, which means power-purchase and utility structures rather than purchase orders.
  • Every dollar clears an appropriations cycle, which pushes meaningful equipment orders into 2027-2028 at the earliest.
  • The statutory demonstrations are the cheapest, fastest test: connecting on-site solar to critical-mission facilities, isolating critical facilities from the commercial grid, and storing renewable energy for night operations.
Defense demand is counter-cyclical, not unlimited. Appropriations, not need, set the pace — and that pace is slow enough to make this a 2027-2028 story rather than a 2026 one.

Supply chain

Who Actually Captures the Dollars

Where the FORTRESS spend lands, and the evidence behind each tier
TierNamesHard evidenceRead
Equipment — generationCaterpillar, Cummins, GeneracCaterpillar revenue $20.68B in Q3 2026, up 17.3%; Cummins Power Systems segment EBITDA margin 48.9% in 1H 2026Prime beneficiaries of a fleet-wide standard
Equipment — switchgear and UPSEaton, Vertiv, Powell IndustriesEaton Electrical Americas backlog $15.2B, up 33%; Vertiv Americas sales up 74.5%; Powell backlog a record $2.4B, up 69%Islanding hardware is the binding constraint
Equipment — fuel cellsBloom EnergyRevenue $1.24B in Q3 2026, up 138%, gross margin 33.7%Fastest growth, most expensive multiple
Equipment — nuclear componentsBWX TechnologiesRevenue $901.6M in the quarter ended June 30, 2026, up 18.0%A 2030s option, not a 2027 order
Integrators — engineering and constructionAECOM, FluorAECOM FY2025 revenue $16.1B with $1.03B operating income; Fluor FY2025 operating loss $363MEquipment, not construction, keeps the margin
Operators — the buyerContinental U.S. installationsAbout 500 bases under assessmentSpend is set by appropriations, not by need

Two exclusions sharpen the map. The first is legal: Section 805 of the National Defense Authorization Act bars procurement from entities designated as Chinese military companies, which keeps Chinese solar, battery and switchgear vendors out of installation contracts and hands that share to allied and domestic suppliers. The second is economic: the nuclear content of FORTRESS is a handful of sites, not a fleet. The Air Force began work on a microreactor at Eielson Air Force Base in 2022, Kadena Air Base in Japan brought on a microgrid with generators and solar in 2023, and two more Space Force bases are candidates for Defense Innovation Unit reactor programs — against roughly 500 bases. Anyone expressing FORTRESS through the reactor complex is buying the rounding error.

  • Winners: generator and switchgear makers with existing base footprints, because a fleet standard repeats one specification 500 times instead of running 500 bespoke projects.
  • Also winners: allied non-US suppliers able to work on US installations, since Chinese vendors cannot bid — Korean turbine and reactor component makers included.
  • Losers: the construction layer, which sees the least of a standardized repeatable package, and the fuel-cell option already trading at 97x trailing EV/EBITDA.

Collision

The Defense Buyer Arrives Into a Sold-Out Market

The resilience mandate is a small problem in megawatts and a large problem in certainty. DLA Energy, the Defense Logistics Agency's power-buying arm, reports managing about 18 million megawatt hours of electricity valued at $1.1 billion under multiple-year contracts — an implied average load of roughly 2 GW. Meanwhile Bloom Energy grew quarterly revenue 138% year over year to $1.24 billion, Powell Industries booked a record $2.4 billion backlog with a 3.0x book-to-bill, and Generac describes data center backup demand as a supply-constrained market. The same equipment lines are already spoken for.

Scale check: the resilience buyer against the commercial buyer
Load or lineScaleRead
DLA Energy electricity contractsAbout 18 million MWh a year, $1.1 billion of valueRoughly 2 GW of average load — the islanding problem is small in megawatts
Bloom Energy quarterly revenue, Q3 2026$1.24 billion, up 138% year over yearOn-site generation is the fastest-scaling answer for the same buyer
Powell Industries backlog, June 2026 quarter$2.4 billion, up 69%, with a 3.0x book-to-billBehind-the-meter generation content is already sold out
Generac C&I segment sales, June quarter$556 million, up 29.2% year over yearBackup power demand is described as supply-constrained
FORTRESS scopeAbout 500 continental basesMany small sites — standardization and funding, not megawatts, are the constraints

That combination has a sharper consequence than a demand estimate. If the lines that serve FORTRESS are already sold out, a 90-day mandate does not automatically flow to the incumbent at the incumbent's price: it flows to whoever can deliver — turbines, fuel cells, second-tier suppliers — and it can also simply wait. The second buyer is a pricing-power question as much as a volume question, and the Department of Energy's own resilience framing, which monetizes islanded microgrids through the value of lost load, treats the whole exercise as an insurance premium on a small, critical load.

What to watch

Horizons: Nothing in 2026, Something in 2027-2028

Two horizons matter, and they are separated by a budget document rather than a product launch.

  • Next 12 months: a costed FORTRESS assessment. Officials said on Oct. 20 they are still working out the cost; the first published per-base cost is the first tradable number.
  • Next 12 months: the fiscal 2027 budget request, where an energy-resilience or installation-infrastructure line with a dollar figure, and military construction items that name microgrids, would convert a memo into a pipeline.
  • Next 12 months: installation-level disclosures at Eielson, Buckley, Malmstrom, Tyndall and Kadena, which will reveal unit costs per base rather than program-wide estimates.
  • One to three years: the statutory demonstration projects under 10 U.S.C. § 2911 — on-site solar feeding critical-mission facilities, critical facilities isolated from the commercial grid, and stored renewable energy for night operations.
  • One to three years: the generator delivery queue. A 90-day standard needs far more fuel and far more standby hours than a 14-day one, landing in a market where data centers already compete for the same production slots.
The thesis fails in one specific way: FORTRESS stays a planning document, or the per-base price proves too high to close. Watch the first published per-base cost, not the base count — the base count is already 500.

Conclusion

The Thesis: A Second Buyer, Late and Cheap

FORTRESS is a genuine, decades-long mandate that converts continuity planning into a repeating order for standardized equipment. It is not yet an order book, and its physical content in megawatts is small. The cleanest expression is the distributed-power and switchgear layer bought at industrial multiples: Cummins at 15.2x and Caterpillar at 17.7x trailing EV/EBITDA against Vertiv at 31.2x, Eaton at 26.9x and Bloom Energy at 97.3x, with Powell Industries at 22.1x on a net cash balance sheet as the small-cap version. The nuclear trade is the wrong side of the arithmetic — BWX Technologies is a 2030s option — and the contractor layer is where the money does not go: Fluor lost $363 million operating money in a year with $15.5 billion of revenue.

Own the lines that already serve hyperscale and let the defense buyer arrive on an appropriations timetable: the second buyer is real, and it is late.

How the mandate reaches each name

CCaterpillarCAT--
--Vol --
-
Bullish
  • Third-quarter revenue of $20.68 billion grew 17.3% year over year; a 500-base islanding standard lands on the same genset lines.
  • Trades at 17.7x trailing EV/EBITDA against Vertiv at 31.2x, the cheapest large-cap way to own distributed power.
  • Near term no FORTRESS revenue exists; the costed assessment and the fiscal 2027 request are the catalysts.
CCumminsCMI--
--Vol --
-
Bullish
  • Power Systems sales rose 15.5% to $1.22 billion in the June quarter, with segment EBITDA margin of 48.9% in the first half.
  • 48.9% segment EBITDA margin leaves room to serve a government buyer without giving up the data center premium.
  • At 15.2x trailing EV/EBITDA it is the cheapest of the genset primes and the most direct beneficiary of a 90-day standard.
  • One to three years out, fleet-wide islanding standardization is the volume driver; near term the growth is data center.
GGeneracGNRC--
--Vol --
-
Bullish
  • C&I sales rose 29.2% year over year to $556 million in the June quarter while Residential fell 2.2%, a clean mix shift to backup power.
  • Management calls data center backup demand supply-constrained; a 90-day standard adds a second, less price-sensitive buyer.
  • Concentrated in C&I, so defense volume lands in the segment already running a 44.5% gross margin.
  • One to three years out this is the highest-beta way to own duration; near term it trades on data center news, not defense.
VVertivVRT--
--Vol --
-
Mixed
  • Americas segment sales of $2.99 billion grew 74.5% year over year, with services and spares adding $719 million in the quarter.
  • Islandable microgrids need exactly the UPS, switchgear and controls it sells, and services attach to every base it commissions.
  • At 31.2x trailing EV/EBITDA, the multiple already prices a data center buyer and leaves less room for the defense option.
  • Direction depends on mix: standardized product holds margin, bespoke installation work dilutes it.
EEatonETN--
--Vol --
-
Bullish
  • Electrical Americas backlog hit $15.2 billion at June 30, 2026, up 33%, with organic orders up 41% and a book-to-bill of 1.3.
  • The segment's organic growth is attributed to data centers and machine OEMs; FORTRESS is the same switchgear specified again.
  • Book-to-bill of 1.3 means the line is already full before any defense volume is added.
  • One to three years out, a standardized islanding specification across 500 sites is the bull case; near term utility work dominates.
PPowell IndustriesPOWL--
--Vol --
-
Bullish
  • Backlog reached a record $2.4 billion in the June quarter, up 69%, with record new orders of $934 million and $634 million of cash.
  • Medium-voltage switchgear is what an islanded microgrid needs at the point of connection, and a fleet standard repeats that specification.
  • Net cash and a 22.1x trailing EV/EBITDA multiple make it the cleanest small-cap way to own base hardening.
  • Near term its backlog is utility, industrial and one data center order above $400 million; defense needs the first 90-day specification.
BBWX TechnologiesBWXT--
--Vol --
-
Watch
  • Revenue of $901.6 million in the quarter ended June 30, 2026 grew 18.0%, driven by naval and commercial nuclear rather than bases.
  • FORTRESS names portable reactors, but the live base count is one microreactor at Eielson plus two Space Force candidates.
  • Nuclear content is a rounding error against 500 bases, which makes this a 2030s option rather than a FORTRESS trade.
  • Watch the Eielson milestone and the Defense Innovation Unit decisions at Buckley and Malmstrom; 25.5x forward earnings already pays for the story.
0Doosan Enerbility034020.KS--
--Vol --
-
Watch
  • One of the few non-US suppliers able to work on US installations, with FY2025 revenue of ₩17.1 trillion across turbines, microgrids and reactor components.
  • Chinese vendors are barred from defense procurement, so allied manufacturers inherit whatever second buyer FORTRESS creates.
  • Operating profit fell to ₩706 billion in FY2025 from ₩1.02 trillion, so turbine volumes are already margin-constrained.
  • One to three years out the opportunity is US behind-the-meter gas generation; near term the catalyst is a US turbine order.